---
title: "Cloud GPUs for Fine-Tuning LLMs — QLoRA from $0 | PowerGPU"
description: "Fine-tuning GPUs at fixed prices: QLoRA an 8B for ~$0, 70B overnight on one 80 GB card, full-parameter on 8× A100. Live prices, axolotl template, walkthrough."
url: https://powergpu.ai/use-cases/fine-tuning
last_modified: 2026-09-14T11:04:20+00:00
prices_as_of: 2026-09-14
site: PowerGPU (powergpu.ai)
---

Use case · fine-tuning

# Fine-tune LLMs on cloud GPUs: your data, an evening's budget

Adapter methods moved fine-tuning from cluster territory to single-card territory: QLoRA a 8B model for about **$0.24** on an interruptible RTX 4090, or a 70B overnight on one 80 GB card. Per-second billing means the meter stops with the last step.

## The fine-tuning ladder

| Tier | GPU | VRAM | On-demand | Interruptible | Why this card |  |
| --- | --- | --- | --- | --- | --- | --- |
| (Good) | [RTX 5090](https://powergpu.ai/gpu/rtx-5090) | 32 GB | $0.439 | $0.219 | 32 GB GDDR7 — QLoRA up to ~34B on one card, twice the memory bandwidth of the 4090. | [Deploy](https://cloud.powergpu.ai/?gpu=rtx-5090) |
| (Better) | [A100 SXM4](https://powergpu.ai/gpu/a100-sxm4) | 80 GB | $0.560 | $0.280 | 80 GB HBM2e — the classic LoRA/full-FT node; 8× SXM4 scales without drama. | [Deploy](https://cloud.powergpu.ai/?gpu=a100-sxm4) |
| (Best) | [H100 PCIE](https://powergpu.ai/gpu/h100-pcie) | 80 GB | $1.867 | $0.933 | 80 GB + FP8 — full fine-tunes finish in roughly half the A100 wall-clock. | [Deploy](https://cloud.powergpu.ai/?gpu=h100-pcie) |

Budget corner: a 24 GB [RTX 4090](https://powergpu.ai/gpu/rtx-4090) at $0.163/hr interruptible handles 8B–13B QLoRA beautifully.

## What fits where (QLoRA, 4-bit base)

| Base model | VRAM needed | Cheapest single card | ≈ per training hour |
| --- | --- | --- | --- |
| Llama 3.1 **8B** | ~11 GB | [RTX 3090 24 GB](https://powergpu.ai/gpu/rtx-3090) | $0.054 |
| Qwen 2.5 **14B** | ~17 GB | [RTX 4090 24 GB](https://powergpu.ai/gpu/rtx-4090) | $0.163 |
| Qwen 2.5 **32B** | ~26 GB | [RTX 5090 32 GB](https://powergpu.ai/gpu/rtx-5090) | $0.219 |
| Llama 3.1 **70B** | ~48 GB | [H100 PCIe 80 GB](https://powergpu.ai/gpu/h100-pcie) | $0.933 |

Interruptible rates shown — fine-tunes checkpoint, so pay half. Full tables in the [VRAM guide](https://powergpu.ai/guides/llm-vram-requirements).

## A clean fine-tuning loop

1. Stage data on a volume Dataset + output dir live on a $0.08/GB/mo volume, not the disposable instance.
2. One YAML, one command axolotl template: axolotl train qlora.yml — checkpoints stream to the volume.
3. Merge, serve, destroy Merge adapters, point a [vLLM instance](https://powergpu.ai/use-cases/llm-inference) at the volume, destroy the trainer. Billing: zero.

## Fine-tuning GPUs: FAQ

The full copy-paste run lives in the [QLoRA walkthrough](https://powergpu.ai/guides/fine-tune-llm-qlora).

**What does a typical fine-tune cost?**

A QLoRA pass over 10k instruction pairs on Llama-3.1-8B takes ~1.5 h on one RTX 4090 — about $0.24 interruptible. A 70B QLoRA on one 80 GB card is an overnight run in the tens of dollars. Full-parameter 8B on 8× A100: low hundreds.

**LoRA, QLoRA or full fine-tune?**

QLoRA first: 4-bit base + trainable adapters fits 8B in under 12 GB and 70B in ~48 GB, and quality is usually within a point of full tuning for instruction tasks. Go full-parameter only when the domain shift is large — then rent A100/H100 nodes for hours, not weeks.

**Which template should I start from?**

axolotl — one YAML covers LoRA/QLoRA/full, DeepSpeed and FSDP, and the walkthrough guide is written against it. Kohya covers the image-LoRA side.

---

*About PowerGPU:* PowerGPU (powergpu.ai) is a cloud GPU rental service offering 80 NVIDIA GPU models — from the RTX A2000 at $0.024/hr to the B300 — at fixed prices set at least 30% below the public GPU marketplace median and re-checked weekly (H100 SXM: $1.428/hr on-demand). Billing is per second with no minimums; payment is crypto only (USDT, BTC, XMR, ETH, SOL, LTC, TRX) with no KYC. Instances run in Tier-III datacenters across 32 regions with a 99.9% uptime SLA and deploy in about 30 seconds from the web console (cloud.powergpu.ai) or the REST API.

Source: https://powergpu.ai/use-cases/fine-tuning · Site index for AI assistants: https://powergpu.ai/llms.txt · Full content: https://powergpu.ai/llms-full.txt